Transcript
Page 1: Design and MIMO control of A Hyper-Redundant Robotic Arm

Design and MIMO Control of A Hyper-Redundant Robotic ArmXingsheng Xu

Advisor: Raul OrdonezDepartment of Electrical and Computer Engineering, University of Dayton

Key Words•Hyper-redundant robots (HRR) are a kind of robots that have a

very large degree of kinematic redundancy.

• Forward kinematics is an arithmetic operation, which determinesthe position and orientation of the end-effector given the values ofeach joint parameter of the robot using the kinematic equations.

•Multi-input multi-output (MIMO) control approach inputs torqueof each joint to control joint dynamic variables such as position, ori-entation, velocity and acceleration in a hyper-redundant robotic sys-tem.

(a) Snake (b) Elephant Trunk (c) Tentacle

Figure 1: Hyper-redundant robots.

Motivation•An application robotic platform has been constructed based on both

the kinematic and dynamic model of a hyper-redundant manipulator.

•Also, we apply the MIMO input-output feedback linearizationcontrol approach to this hyper-redundant robotic system to controlthe end-effector in Cartesian space. This control approach can highlyimprove the robotic performance while executing motion control ortracking a path in a constrained and complex environment.

Kinematic Model

(a) Joint schematic (b) Frame assignment

This model is established at its home position. The arm is composedof a set of revolute joints which are arranged orthogonal to each otherwhile making the shoulder and wrist joint spherical.

Performance without Dynamic Control• Tracking A Rectangle on X-Y Plane:

Figure 2: Tracking-trajectory of a rectangle on X-Y plane.

Figure 3: Velocity of tracking-trajectory of a rectangle on X-Y plane.

• Tracking An Inclined Circle:

Figure 4: Tracking-trajectory of an inclined circle.

Figure 5: Velocity of tracking-trajectory of an inclined circle.

Dynamic Model•Manipulator Jacobian Matrices: An expressions to connect the re-

lationship between angular velocity ω0n, linear velocity v0

n of the end-effector and joint velocity q as

ω0n = Jωq,

v0n = Jvq,

where Jω and Jv are 3× n matrices.

• Euler-Lagrange Equation: An application leads to a robotic systemof n coupled, second order nonlinear ordinary differential equationsof the form

d

dt

∂L

∂qi− ∂L

∂qi= τi, i = 1, ..., n,

where τi is input torque of each motor and the the Lagrangian L isgiven by

L = K − P,

where K is the kinetic energy and P is the potential energy.

Algorithm of MIMO Control• Joint Space MIMO Control: We can rewrite the system dynamics

from the Euler-Lagrange equation into Input-Output form as

qi = f (qi, qi) + g(qi, qi) · τi,

and outputs

yi = qi, i = 1, ..., n.

Then, we pick our MIMO controllers as

τi = g(qi, qi)−1 · [−f (qi, qi) + υi], i = 1, ..., n.

The output tracking errors will be

ei = qi − ri,

where ri is joint variable reference. We set υi = −kiei − ki+1ei andget outputs yi = −kiei − ki+1ei.

•Cartesian Space MIMO Control: The Jacobian matrix can be asmooth and invertible mapping between the joint variables q ∈ Qand the workspace variables Γ ∈ Rn as

Γ = Jq.

We take the derivative of both sides of the equation, then substitute qcreated by the joint space to get

Γ = F (qi, qi) + G(qi, qi) · τi,

where F (qi, qi) = J · f (qi, qi) + J q and G(qi, qi) = J · g(qi, qi).After that, we can derive our MIMO workspace controllers as

τi = G(qi, qi)−1 · [−F (qi, qi) + υi], i = 1, ..., n.

The output tracking errors will be

ei = Γi − ri,

where ri is end-effector variable reference. We set υi = −kiei −ki+1ei and get outputs yi = −kiei − ki+1ei.

Performance Applying MIMO Control• Tracking A Circle Reference Path on X-Y Plane:

• Input Controllers:

• System Outputs:

Extra Constrains of Redundant System•Mathematical Requirement:

The 6 × n matrix G(qi, qi) should be a square matrix in order toderive the Cartesian space MIMO controllers. Therefore, we needto find n − 6 more constrain equations for a redundant system (ie.n > 6).

•Additional Constrains:The constrain equations can be chosen differently according to theobjective of the redundant robotic arm. One can pick geometric lim-itations of specific joints to avoid obstacles or motion optimizationsof the whole arm to save energy, etc.

Top Related